A coupled plant simulation system TRAC/BF1-ENTRÉE was applied to the Nuclear Energy Agency/National Security Council boiling water reactor turbine trip benchmark. Through regular exercise 3 and extreme scenarios 3 and 4, its adequacy and robustness were validated. It was deduced that the cross-section format and the core boundary conditions are major influential factors causing errors in three-dimensional power predictions. Power swings observed in extreme scenarios were attributed to intermittent void generation and void sweeping driven by rapid pressurization. Based on a series of sensitivity studies for extreme scenario 4, it was confirmed that neglect of in-channel direct heating causes a large positive reactivity insertion and neglect of bypass direct heating causes only a small change in reactivity effects. Specifying an integration time-step size of <1 ms is recommended for keeping the numerical error within an acceptable level. To investigate the detailed in-channel void distribution and its possible influences on the fuel thermal margin, a one-way coupled system between TRAC/BF1-ENTRÉE and the three-field subchannel code NASCA was developed. Detailed void distributions at the upper part of the core where the boiling transition will occur become sufficiently uniform during the major period of the turbine trip event. Their influences on the thermal margin seem negligible.